Path: news.cs.au.dk!news.net.uni-c.dk!uninett.no!news-feed.inet.tele.dk!bofh.vszbr.cz!newspeer.monmouth.com!nntp2.dejanews.com!nnrp1.dejanews.com!not-for-mail
From: Andrew Klaassen <clawsoon@my-dejanews.com>
Newsgroups: comp.lang.beta
Subject: Re: Virtual attribute gives Seg Fault when accessed from pattern ref
Date: Mon, 03 May 1999 23:47:48 GMT
Organization: Deja News - The Leader in Internet Discussion
Lines: 94
Message-ID: <7glcj3$tu7$1@nnrp1.dejanews.com>
References: <7gi6gi$99i$1@nnrp1.dejanews.com> <372D57DD.41C6@cs.au.dk> <7gkb3t$vho$1@nnrp1.dejanews.com> <yc0emkymazo.fsf@ufleku.cs.au.dk>
NNTP-Posting-Host: 207.34.91.201
X-Article-Creation-Date: Mon May 03 23:47:48 1999 GMT
X-Http-User-Agent: Mozilla/3.04 (X11; I; Linux 2.2.4 i486)
X-Http-Proxy: 1.0 x3.dejanews.com:80 (Squid/1.1.22) for client 207.34.91.201
Xref: news.cs.au.dk comp.lang.beta:11915

In article <yc0emkymazo.fsf@ufleku.cs.au.dk>,
  nospam2159@cs.au.dk (Peter von der Ahé) wrote:
> >>>>> "AK" == Andrew Klaassen <clawsoon@my-dejanews.com> writes:
>
>  AK> ...  Function: (# ... Printname: (#...#); do ... #); ...  do
>  AK> Function.Printname; (* doesn't compile *) ...
>
>  AK> Why not?
>
> This will create a new instance of the Pattern "Function.Printname".
> Do to that the Function needs to be an object, otherwise Printname has
> no origin or surrounding environment.
>
> "(&Function).Printname" should work.  This uses a computed reference.
>
> A computed reference has the form:
>
>    (EXPRESSION).ATTRIBUTE
>
> where EXPRESSION must return a reference qualified by something which
> has an attribute ATTRIBUTE.
>
  ---snip---

(&Function[]).Printname does the trick.

>
> Another thing to consider is efficiency,

I'm not too concerned about efficiency in this particular
case - in the actual code I'm working on, the equivalent of
"Printname" will only be used for debugging.  If computed
referencing makes things more convenient (and I'm still
deciding if it will), I'll ignore the performance loss.

However, code speed might be important in regard to some of
my other functions.  Something like, say

IntegerPartOfLog2:
  (# value,i: @integer;
     enter value
     do
       ...
     exit (i - 1)
  #);

Normally, I'd call this like so:

i,a: @integer;
do
  (for i:30000 repeat i -> IntegerPartOfLog2 -> a; ... for);

(Don't worry - in the actual code, I use the function
far more intelligently than this<G>.)

Would my code be faster overall (since the function
will be called many times, and since I never need more
than one copy of it) if I made IntegerPartOfLog2 a
singular object rather than a pattern? (I guess that
that would depend on whether i->IntegerPartOfLog2
creates a new instance of the pattern each time it's
executed or not.)

> if you intend that Function
> should be a collection of some methods used for similar things,
   ---snip---

Actually, Function is intended to be a superpattern for a
number of methods that are to be used in the same place
and are referenceable (is that a word??) from the same
array, as in

Function: (# ... PrintName:< (# do inner #) do inner #);
F1: Function
 (# ... PrintName:: (# do 'F1' -> puttext; #) ... do ... #);
F2: Function
 (# ... PrintName:: (# do 'F2' -> puttext; #) ... do ... #);
(* etc.*)
 (* I might also do F1: @Function(#...#) - I haven't decided
  * yet. There are a few considerations that are still a bit
  * fuzzy in my mind.
  *)
FunctionArray: [0] ##Function;
(* or *)
FunctionArray: [0] ^Function;

> Cheers,
> Peter

Thanks,
Andrew

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own